Harmonic oscillator transition kernel (source code)

= Harmonic oscillator transition kernel
{title2=$K_\omega(q_f,q_i;T)$}

The <harmonic oscillator transition kernel> is the position-space <unitary time evolution> amplitude
$$
K_\omega(q_f,q_i;T)=\langle q_f|e^{-iHT/\hbar}|q_i\rangle,\qquad H=\frac{p^2}{2m}+\frac12m\omega^2q^2.
$$
For $0<T<\pi/\omega$,
$$
K_\omega=\sqrt{\frac{m\omega}{2\pi i\hbar\sin(\omega T)}}\exp\left\{\frac{im\omega}{2\hbar\sin(\omega T)}\left[(q_f^2+q_i^2)\cos(\omega T)-2q_fq_i\right]\right\}.
$$
Here $\hbar$ is the <reduced Planck constant>. The interval fixes the initial square-root branch; extension beyond focusing times requires its continued phase.

In the <path integral>, write a path as the classical oscillator path plus a fluctuation vanishing at both endpoints. The quadratic <action> splits exactly into the classical <action> and the endpoint-independent fluctuation <action>. The <Dirichlet oscillator determinant ratio> gives the prefactor, while the classical boundary <action> gives the exponent. <The Path Integral approach to Quantum Mechanics by Riccardo Rattazzi>[Printed pp. 23–26, equations 1.75–1.99, especially determinant factor 1.98 and kernel 1.99]{ref}

As $\omega\to0$ with $T$ fixed, the expression reduces to the <free-particle propagator>
$$
K_0=\sqrt{\frac m{2\pi i\hbar T}}\exp\left[\frac{im(q_f-q_i)^2}{2\hbar T}\right].
$$
At $T=n\pi/\omega$, the coordinate kernel is distributional because <classical paths> refocus. The <Feynman i-epsilon prescription> fixes the continuation through these caustics. Integrating the kernel against an initial <wavefunction> gives its later <wavefunction>.